For an AI loser, Apple did an awful lot of winning last year.
The mess that was the Apple Intelligence rollout was embarrassing, to be sure, but through it all, the company kept doing what it does best: selling iPhones. With this week’s news that it’ll use Gemini models to power the long-awaited smarter Siri, Apple seems to have taken a big ‘ol L in the whole AI race. But there’s still a major challenge ahead — and Apple isn’t out of the running just yet.
Apple Intelligence got off to a well-documented rough start in 2024. The iPhone 16 was “Built for Apple Intelligence,” but shipped without it. Features arrived over the next few months, but the so-called smarter Siri never materialized. Apple execs admitted they had gone back to the drawing board, people in charge were shuffled around, and it all looked like a big failure on Apple’s part.
But it doesn’t exactly seem like people are ready to ditch their iPhones for Google’s Gemini-imbued Android phones. According to IDC’s Q3 2025 report, “Demand for Apple’s new iPhone 17 lineup was robust, with pre-orders surpassing those of the previous generation.” Counterpoint Research calls Apple the global smartphone “market leader” in 2025 with 10 percent year-over-year growth in market share. Meanwhile, Apple Intelligence is much less prominent in the iPhone 17’s marketing than it was with the 16; you have to scroll halfway down the iPhone 17’s product page before you get to the first mention.
The stalling tactic worked, but these days investors break out in hives if you don’t mention AI every five minutes. Apple had to come up with something in the way of a strategy, and in the second half of 2025, we started hearing reports that it might be looking at outside partners, rather than building its own models from scratch. It wouldn’t be totally unprecedented; Apple already lets users access ChatGPT directly in iOS and has promised from the beginning that it would add more third-party LLMs in this fashion. But this week’s deal isn’t about adding a quick way to chat with Gemini on an iPhone. You can already do that in the Gemini app. This is about building smarter Siri on Google’s models and running it all in Apple’s Private Cloud Compute. If and when a smarter Siri arrives this year, it’ll have some serious Gemini DNA.
You could argue Apple made the right call from a business standpoint, but was it the right Apple move? Consider Tim Cook’s own words in a 2009 earnings call: “We believe that we need to own and control the primary technologies behind the products we make…” That was the foundation of the company’s push to develop its own silicon, which has absolutely been a winning strategy. But either Apple thinks AI models aren’t a primary technology after all — more like an underlying service it will build products on top of — or it has made a serious misjudgment about AI as the next platform shift and risks falling behind. Low stakes stuff!
That’s the challenge: turning Apple Intelligence into a product people actually want, not one that they feel indifferent toward
And to be sure, Apple doesn’t control the destiny of every part of the iPhone. It hasn’t gone out and built a search engine, or its own wireless network, or an algorithmic social media platform. All of those things run on an iPhone, but they aren’t core parts of the iPhone’s identity, and AI could very well wind up the same way. Maybe there’s a hint in the way that Apple has seemingly shifted from encouraging developers to adopt its own App Intents framework to using the Anthropic-developed MCP as the basis for agentic features. If AI just needs to find the right hooks to get things done, then the particular models it’s running matter less. But it all hinges on the product that Apple creates around AI — and that starts with Siri.
That’s the real challenge ahead of Apple: turning Apple Intelligence into a product people actually want, not one that they feel indifferent toward. It needs to turn Siri into the thing the company has promised all along, not a glorified timer-setting machine. Apple can make a beautiful product — no doubt. Can it do that without control over its own models? Can it do that faster than Google or Jony Ive or any other competitor ready to make a run at the garden walls? The deal may be inked, but the real work starts now.